1. Field of the Invention
This invention relates to a machine visual apparatus, and more particularly to an automatic inspection system using barcode locator and a method thereof.
2. Description of Prior Art
Nowadays, machine vision has been widely adopted in the automatic inspection field, by which cameras and computers, instead of the human vision, are used to perform estimation and inspection, etc. It is accurate, reproducible and high-speed. Using machine vision for automatic inspection can reduce the cost and improve the quality of the products.
A typical application of machine vision is the inspection of PC boards, which is an operation of one-dimensional inspection for a two dimensional plane. Many high-speed inspections are needed while estimating the size of inductance and the intactness of the circuit. Machine vision can provide a high accuracy and reliability that is difficult to achieve by human inspection.
In U.S. Pat. No. 5,546,475, Bolle et al. disclose a produce recognition system which uses the histogram technique as the basis of segmenting and feature extraction. In U.S. Pat. No. 5,465,308, the input image is recognized through spectrum analysis and neural networks.
The above prior arts have to perform a large number of correlation operations. That is, an excellent processing speed must be provided; otherwise, it will take a long time to perform the operation. Therefore, U.S. Pat. No. 5,495,537, xe2x80x9cMethods and apparatus for machine vision template matching of image predominantly having generally diagonal and elongate featuresxe2x80x9d, introduces the operations of hill climbing and segment peak finding into the 4-way or 8-way search system to reduce the number of correlation operation.
When recognizing an object, the brightness distribution on the surface of the object has a significant effect on the recognition result. To avoid the difficulty of object recognition at different viewing angles due to various illuminations, U.S. Pat. No. 57,227,080 discloses a technique similar to the dynamic time warping method to perform dynamic histogram warping of the image histograms, thereby achieving constant image brightness for stereo pair images.
A typical on-line application of machine vision uses the edge detection technique to enhance the characteristics and performs several certain measurements to filter out defective products. Basically, it is a one-dimensional detection technique, which is not suitable for two-dimensional inspection of the surface of a three-dimensional object.
The current inspection techniques for three-dimensional objects, such as aluminum cans, is not accurate enough for the product line. Furthermore, it cannot proofread a whole two-dimensional surface.
Accordingly, to overcome the drawbacks of the prior arts, the object of this invention is to provide an automatic inspection system using barcode locator and a picture-content proofreading technique, which can provide a high-speed, high accuracy inspection for a three-dimensional object.
The automatic inspection system of this invention has two features: First, it is a three-dimensional inspection system. Taking the inspection of an aluminum can as an example, the on-line three-dimensional inspection is accomplished with uncertain view angles. Second, it provides the ability to rapidly switch the type of objects inspected. That is, the inspection system can inspect a variety of different cans if their outlines are similar to each other.
For the automatic inspection of aluminum cans, the direction of aluminum can is not pre-determined when the image of the aluminum is retrieved on the production line. Thus, the reference point has to be located before comparison. Nowadays, every aluminum can is provided with a barcode, which has a significant area and is easy to recognize. Therefore, the barcode is used as a reference position in this invention.
To achieve the above object, this invention provides an automatic inspection system using barcode locator to inspect a colored surface of a three-dimensional object having a barcode. The automatic inspection system comprises: a light source for uniformly illuminating the three-dimensional object; an image retrieving device for retrieving and analyzing images of the object from various viewing angles; an operating and processing device for receiving data of three- dimensional images output from the image retrieving device, segmenting the three-dimensional images, converting the three-dimensional images into two-dimensional images, then connecting the two-dimensional images of various viewing angles to obtain an image representing the three-dimensional object, and comparing the two-dimensional image with a reference object image previously stored in the operating and processing device to find defects of the inspected object.